Evolution in the Understorey: the Sulawesi Babbler Pellorneum Celebense

Home , Pellorneidae, Pellorneum, Sulawesi babbler, Sylviidae

Zoologischer Anzeiger 293 (2021) 314e325

Contents lists available at ScienceDirect

Zoologischer Anzeiger

journal homepage: www.elsevier.com/locate/jcz

Evolution in the understorey: The Sulawesi babbler Pellorneum celebense (Passeriformes: Pellorneidae) has diverged rapidly on land- bridge islands in the Wallacean biodiversity hotspot * Fionn O Marcaigh a, , David J. Kelly a, Darren P. O'Connell a, b, c, Daniel Dunleavy a, Alice Clark a, Naomi Lawless a, Adi Karya d, Kangkuso Analuddin d, Nicola M. Marples a a Department of Zoology, School of Natural Sciences, Trinity College Dublin, Dublin, D02 CX56, Ireland b School of Natural and Environmental Sciences, Newcastle University, Newcastle Upon Tyne, NE1 7RU, UK c School of Biology and Environmental Science, University College Dublin, Dublin, D04 N2E5, Ireland d Department of Biology and Biotechnology, Universitas Halu Oleo, Kendari, South-east Sulawesi, Indonesia article info abstract

Article history: Tropical islands hold great treasures of Earth's biodiversity, but these fragile ecosystems may be lost Received 8 March 2021 before their diversity is fully catalogued or the evolutionary processes that birthed it are understood. We Received in revised form ran comparative analyses on the ND2 and ND3 mitochondrial genes of the Sulawesi babbler Pellorneum 5 July 2021 celebense, an understorey bird endemic to Sulawesi and its continental islands, along with its Accepted 7 July 2021 morphology and song. Genetic, acoustic, and morphological data agree on multiple isolated populations, Available online 8 July 2021 likely representing independently evolving lineages. The Sulawesi babbler shows signs of rapid specia- Corresponding editor: Alexander Kupfer tion, with populations diverging between Central and Southeast Sulawesi, and even on land-bridge islands which were connected within the last few tens of thousands of years. The genetic divergence Keywords: between Sulawesi babbler populations in this time has been around 33% of their divergence from sister Cryptic biodiversity species which have been isolated from Sulawesi for millions of years. This is likely facilitated by the Incipient speciation Sulawesi babbler's understorey lifestyle, which inhibits gene ﬂow and promotes speciation. Similar Island biogeography patterns of endemism are seen in Sulawesi's mammals and amphibians. This work highlights the un- Pellorneidae documented biodiversity of a threatened hotspot, wrought by complex processes of speciation which Integrative taxonomy interact with ecology and geology. Subspeciﬁc taxonomy has at times been controversial, but we argue Wallacea that discrete populations such as these play a key role in evolution. Lying as they do at the heart of the biodiversity hotspot of Wallacea, these islands can reveal much about the evolution of biodiversity at all of its levels, from the gene to the ecosystem. © 2021 The Authors. Published by Elsevier GmbH. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

1. Introduction Wallace (1880) noted that islands vary in their geographical isolation, with important consequences for evolution. Some, which they The 20th century Biological Species Concept cemented the termed oceanic islands, were created by volcanic eruptions or uplift importance of isolation in evolutionary biology, with pivotal works of coral far out at sea. Oceanic islands are separated from continents like those of Mayr (1942, 1959) and Dobzhansky (1937, 1940) and other large landmasses by deep seas, and thus have never been showing that populations become species when they are isolated, connected to them by land. Continental land-bridge islands, on the ﬁrst geographically and then reproductively. The role of geographic other hand, are formed from parts of the continental shelf, and so at isolation in speciation continues to inspire debate, particularly in times of reduced sea level they were a continuous part of the the evolutionary marvels that are the world's islands (e.g. Flantua mainland. Though later work has elaborated on this original clas- et al., 2020; Itescu et al., 2020). Early biogeographers such as siﬁcation (Ali, 2017, 2018), much research on island speciation still focuses on highly isolated oceanic islands such as Hawaii and the Galapagos (Whittaker et al., 2017). A smaller, but growing, body of literature has drawn important evolutionary conclusions from the * Corresponding author. faunas of land-bridge islands (e.g. Lister, 1989; Vartanyan et al., E-mail address: [email protected] (F. O Marcaigh). https://doi.org/10.1016/j.jcz.2021.07.006 0044-5231/© 2021 The Authors. Published by Elsevier GmbH. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). F. O Marcaigh, D.J. Kelly, D.P. O'Connell et al. Zoologischer Anzeiger 293 (2021) 314e325

1993; Keogh et al., 2005; Itescu et al., 2020), which are numerically with an adjoining microcontinent to form Buton (or Butung) Island more numerous, ecologically more complex, and zoologically richer (Satyana & Purwaningsih, 2011). The seas between these five than are oceanic islands (Meiri, 2017). As land-bridge islands have islands are both shallow and narrow, and they were connected by been isolated only briefly, these studies provide evidence of evo- land within the last 20,000 years (Nugraha & Hall, 2018). lution which has taken place at a rapid rate. The diversity of islands and species found in Sulawesi make it an Isolation of an island population results from features of the ideal place to study evolutionary divergence across islands. As well species as well as the island, as some organisms are more likely to as the land-bridge islands described here, there are more isolated maintain gene flow across water barriers than others. Thus, studies islands, including the Sula group and the Wakatobi (or Tukangbesi) of evolution on land-bridge islands tend to focus on particularly archipelago, which have never been connected to Sulawesi by land weak dispersers, such as terrestrial mammals and reptiles (e.g. (Nugraha & Hall, 2018). Descriptions of endemic species from these Lister, 1989; Vartanyan et al., 1993; Keogh et al., 2005). However, more isolated islands indicate that birds readily speciate when even among birds and other strongly dispersing animals, certain separated by such permanent barriers (Kelly et al., 2014; O'Connell ecological and behavioural traits will inhibit gene flow across bar- et al., 2019b; Rheindt et al., 2020). The Sulawesi babbler is found on riers which the organism should be physically capable of crossing all the land-bridge islands of Southeast Sulawesi, making it an ideal (Harris & Reed, 2002). Some birds that fly long distances over land candidate for studies of evolutionary divergence on this shorter will not cross even narrow bodies of water (Diamond, 1981), and time scale. As the populations on Sulawesi, Buton, Kabaena, Muna, even the dispersive species that do colonise remote islands can and Wawonii have been separated for an evolutionarily brief period develop “behavioural flightlessness” in these isolated populations of time, any divergence between them is evidence of evolution (Moyle et al., 2009). The habit of foraging in forest understorey occurring at a rapid pace. These islands remain poorly known seems to be particularly significant in limiting gene flow and ornithologically, with species inventories emerging only recently driving speciation (Burney & Brumfield, 2009; Smith et al., 2014). (Martin et al., 2012; Martin et al., 2015; Martin et al., 2017; This study aims to investigate evolution on the land-bridge islands O'Connell et al., 2017; O'Connell et al., 2019a). As pressures on surrounding Sulawesi using a bird which is limited to the forest Indonesia's birds continue to mount (Rentschlar et al., 2018), it understorey, the endemic Sulawesi babbler Pellorneum celebense. becomes increasingly urgent that we study recently documented This species was first described, as Trichastoma celebense, by populations such as these, in order to estimate their evolutionary Strickland (1849). The genus Trichastoma (Blyth, 1842) was sub- distinctiveness and consider their conservation. sumed into Pellorneum (Swainson, 1831)byMoyle et al. (2012) and The Sulawesi babbler belongs to the family Pellorneidae (ground Cai et al. (2019). Our taxonomy follows Gill et al. (2021). babblers), which was formed after species were split from the Sulawesi is the largest island of the Wallacea region, one of Timaliidae, Sylviidae, and Cisticolidae on molecular evidence (Cai Earth's threatened biodiversity hotspots (Myers et al., 2000). et al., 2019). Babbler systematics are complex, and scientists' in- Sulawesi's complex geology has shaped an “anomalous” biogeog- terpretations of them have changed repeatedly over the years raphy (Wallace, 1880) and a high level of endemism (Stattersfield (Cibois et al., 2002; Cibois, 2003; Gelang et al., 2009; Moyle et al., et al., 1998), with the island divided into four distinct peninsulae 2012; Cai et al., 2019). Pellorneid species richness reflects their or “arms”, here referred to as North, Central, South, and Southeast biogeographic history (Cai et al., 2020), being highest in the Sino- Sulawesi (Fig. 1) (the centre of the island, between the arms, also Himalayan Mountains (where the group originated) and in the forms part of Central Sulawesi). Southeast Sulawesi and the islands Sundaland region. Though Wallacea lies immediately to Sunda- of Kabaena, Muna, and Wawonii (or Wowoni) all sit on a small land's east, babblers colonised it more recently and thus their fragment of continental lithosphere (Hall, 2013), which collided species richness is lower in this region. Aspects of the natural

Fig. 1. a) Map of Indonesia with study region outlined. b) Map of Southeast Sulawesi with sampling sites marked with circles. c) Map of Sulawesi with song recording locations marked with triangles.

315 F. O Marcaigh, D.J. Kelly, D.P. O'Connell et al. Zoologischer Anzeiger 293 (2021) 314e325

history and ecology of babblers remain enigmatic (O Marcaigh Kabaena, 6 for Muna, and 22 for Buton. Birds were also sampled for et al., 2020). As the Sulawesi babbler is “smallish [and] relatively genetic analysis by taking a small number of contour feathers from featureless” (Billerman et al., 2020) and tends to skulk in the the flank. Compared to other feather tracts, sampling of contour understorey, it is of the kind of bird most likely to be overlooked feathers minimises the risk of injury to the bird and avoids dis- (Diamond, 1985; Gaston & Blackburn, 1994). Its “mouse-like” rupting its flight ability and any plumage-based visual signals behaviour (Billerman et al., 2020) suggests that, biogeographically, (McDonald & Griffith, 2011). Genetic sample size was 5 for South- such understorey birds may have more in common with terrestrial east Sulawesi, 1 for Central Sulawesi, 8 for Kabaena, 6 for Muna, 5 mammals than more dispersive species. for Buton, and 1 for Wawonii. We recorded the babbler songs using If we are to understand how populations become species, it is a Zoom H2 Handy Recorder, with a Sennheiser Me62 external natural that we must study populations as well as species. Variation microphone and a Telinga V2 parabolic reflector, and downloaded below the species level provides the raw material for natural se- additional recordings from the website xeno-canto (https://www. lection, as populations will begin to diverge before they evolve xeno-canto.org/). The combined acoustic sample size from both physiological barriers to reproduction (Dobzhansky, 1940). There- sources was 15 for Southeast Sulawesi, 2 for Central Sulawesi, 12 for fore, targeting the species level and below allows us to study both North Sulawesi, 2 for Togian, 17 for Kabaena, and 16 for Buton. Our current and past speciation (e.g. Brelsford & Irwin, 2009; Everson acoustic sampling thus included three of the four Sulawesi babbler et al., 2018). Units of diversity below the species level have been subspecies recognised by Gill et al. (2021). These are the North defined differently through scientific history, as varieties (Linnaeus, Sulawesi subspecies P. c. celebense (Strickland, 1849), the Togian 1766), subspecies (Esper, 1781; Mayr, 1963), incipient species subspecies P. c. togianense (Voous, 1952), and P. c. rufofuscum which (Dobzhansky & Pavlovsky, 1967), conservation units (Coates et al., supposedly covers all of Central and Southeast Sulawesi and the 2018) or Evolutionarily Significant Units (ESUs) (Moritz, 1994). land-bridge islands. Previous taxonomic treatments had a subspe- Though division of subspecific diversity into units has often cies endemic to Southeast Sulawesi, named sordidum by generated controversy, it is evident that this diversity is pivotal to Stresemann (1938) and renamed improbatum by Deignan (1964) as evolution (O'Brien & Mayr, 1991; Phillimore & Owens, 2006). there was already a sordidum subspecies in the same genus. How- Indeed, while the species features in the title of evolutionary bio- ever, White & Bruce (1986) merged improbatum into rufofuscum logy's founding text (Darwin, 1859), the subtitle refers to subspe- based on their similar flank colour, and this move has been retained cific “races”. The Convention on Biological Diversity (1992) by Gill et al. (2021). The fourth subspecies, not covered by our recognises biodiversity at the levels of genes, species, and ecosys- sampling, is P. c. finschi (Walden, 1876), endemic to South Sulawesi. tems, but conservation at the gene level is hampered by lack of data, particularly in the biodiverse tropics (Bickford et al., 2007). Around 2.2. Genetic analyses 2% of vertebrate species are endemic to Wallacea (Myers et al., 2000), making it an urgent conservation priority that we under- It has been demonstrated that divergence in mitochondrial stand Sulawesi's endemism, both above and below the species genes correlates with speciation rate in tropical birds (Harvey et al., level. 2017). We used the mitochondrial DNA (mtDNA) genes NADH de- The Sulawesi babbler provides one example of a bird species hydrogenase subunits 2 and 3 (hereafter ND2 and ND3), to inves- divided into subspecies based on plumage and other typological tigate population genetics of babblers. Barcoding approaches with characteristics, where genetic data have been lacking. Current mtDNA have proved successful in species delimitation (Hebert taxonomy (Gill et al., 2021) assigns all populations from Central and et al., 2004; Kerr et al., 2007; Hebert et al., 2016). While some Southeast Sulawesi, as well as the land-bridge islands, to a single evolutionary histories inferred from mtDNA differ from those subspecies P. c. rufofuscum (Stresemann, 1931). The wide range thus inferred from nuclear DNA (Rubinoff & Holland, 2005; Phillimore attributed to this subspecies crosses several present-day mountain et al., 2008), and biogeographic patterns inferred from mtDNA ranges and seas. Central and Southeast Sulawesi were separate can be obscured by introgression and male-mediated gene flow islands in the past, leading to differences between their monkeys, (Toews & Brelsford, 2012), ND2 has shown a particularly high level toads (Evans et al., 2003), hoofed mammals (Frantz et al., 2018), and of concordance with nuclear markers (Campillo et al., 2019), and trees (Trethowan et al., 2020). If flight allowed all birds to transcend studies on young radiations have found ND2 and ND3 to provide this pattern, we would expect the widely distributed babblers the best phylogenetic resolution (Andersen et al., 2015). This makes designated as rufofuscum to present one genetically uniform pop- them appropriate to study divergence at the level of populations ulation, following the current taxonomy. Alternatively, if the and subspecies, as we aimed to do. understorey lifestyle of babblers limited their gene flow in a DNA was extracted from feathers using a Qiagen DNeasy Blood manner similar to land mammals, we would expect to see diver- and Tissue Kit (Qiagen, California, USA), following the manufac- sification between Central and Southeast Sulawesi populations and turer's instructions, but with the addition of 5 mL of 1M dithio- potentially across land-bridge islands as well. threitol (DTT). The DTT was added before the samples were vortexed prior to incubation, to break down the keratin from the 2. Materials and methods base of the feather, which encased the genetic material. Polymerase Chain Reactions (PCRs) were carried out in 20 ml reactions to target 2.1. Sampling and data collection the ND2 and ND3 genes, using a touchdown cycling protocol to increase yield (Korbie & Mattick, 2008). We amplified the ND3 gene We mist netted birds in Southeast Sulawesi (Fig. 1, Table S1) and using the L10755-F and DOC-ND3-R1 primer pair (Chesser, 1999; the land-bridge islands of Kabaena, Muna, Buton, and Wawonii, O'Connell et al., 2019b), while ND2 was sequenced in two halves between 1999 and 2017. We used the methods outlined in Redfern using established and novel internal and external primers & Clark (2001) to photograph our birds (Fig. S1) and measure a (Table S2). The reactions were screened using 2% electrophoresis range of morphological traits including wing length (maximum gels stained with GelRed (Biotium), then sequenced by GATC chord), bill length (tip of bill to the base), and skull length (from EuroFins using a Sanger sequencing protocol. These sequences back of skull to base of bill). Only adult birds measured by a single were aligned using the ClustalW function in BioEdit (Hall, 1999) recorder (NMM) were used for morphological analyses. This gave a and the ND2 and ND3 sequences were concatenated using morphological sample size of 22 for Southeast Sulawesi, 18 for Mesquite (Maddison & Maddison, 2018), for a total of 1392bp. We

316 F. O Marcaigh, D.J. Kelly, D.P. O'Connell et al. Zoologischer Anzeiger 293 (2021) 314e325 used GenBank to obtain the sequences of the sole individual of maximum intraspecific divergences, and for each of them it cal- P. celebense that had been sequenced previously (ND2 ccession culates a “barcode gap” which is equal to the minimum threshold JN826691, ND3 accession JN826966), and those of its two sister interspecific distance. It then splits the sequences into groups species: Pellorneum rostratum, described by Blyth (1842) (ND2 separated by the barcode gap. The range of prior maximum intra- JN826692, ND3 JN826967), and Pellorneum bicolor, described by specific divergences allows the analysis to be calibrated for Lesson (1839) (ND2 JN826690, ND3 JN826965). These sequences of different genes and species. We ran ABGD analysis on the web- the three Pellorneum species were published in Moyle et al. (2012), server https://bioinfo.mnhn.fr/abi/public/abgd/ using default set- with the P. celebense material supplied by a bird from Banggai tings (Pmin ¼ 0.001, Pmax ¼ 0.1, Steps ¼ 10, relative gap Province in Central Sulawesi. Our outgroup included other babbler width ¼ 1.5, Number of bins ¼ 20) and a Kimura-2-Parameter (K2P) genera (Cai et al., 2019) and outgroup taxa used by Moyle et al. model. (2012) (a full list with accession numbers is in Table S3). Our new ND2 and ND3 sequences have been deposited in GenBank under 2.3. Phylogeographic analysis accession numbers MW387438- MW387487. We used POPART (Leigh et al., 2015) to draw a TCS Network of Phylogeographic analyses were carried out in R version 4.0.2 (R the haplotypes we sequenced (Fig. 2), to help visualise any poten- Core Team, 2020) using the package Geneland (Guillot et al., tial population structure. The TCS algorithm uses an agglomerative 2005a). This is one of several clustering algorithms useful in approach, progressively combining clusters with one or more assigning genetic data to groups without prior knowledge (Carstens connecting edge (Templeton et al., 1992). A complete list of samples et al., 2013). These algorithms search for the number of genetic and their corresponding haplotypes is available in the Supple- populations that maximises HardyeWeinberg Equilibrium (HWE) mentary Information (Table S3). Only one representative of each and Linkage Equilibrium. Because of this basis in HWE, the package ND2/ND3 haplotype was included in the Maximum Likelihood and creators advise caution in the interpretation of clustering in non- Bayesian analyses. We used MEGA X (Kumar et al., 2018)to recombining DNA, such as mtDNA (The Geneland Development generate pairwise proportion differences (p-distances) between Group, 2020). Nevertheless, previous studies using the Geneland our concatenated ND2/ND3 haplotypes (Table S4) and to choose a package have found clustering of the ND2 gene to agree with that of nucleotide substitution model based on the Bayesian Information nuclear DNA (Trier et al., 2014; Klein et al., 2016). We ran a spatial Criterion (BIC). It selected a Hasegawa-Kishino-Yano (HKY) model, MCMC algorithm on the 51 polymorphic sites of our concatenated which we used to perform Maximum Likelihood analysis with 1000 babbler sequences for 45,000 iterations with thinning set to 100. bootstraps and a level 5 Subtree-Pruning-Regrafting heuristic. We We used Geneland's “haploid” setting as our sequences were of carried out Bayesian phylogenetic inference in MrBayes version mtDNA. This model's output was then used to estimate the number 3.2.7 (Huelsenbeck & Ronquist, 2001), using a HKY model with of populations at HWE and assign each individual to a population burn in set to 25%. This consisted of two independent Markov chain (Fig. 4). The Geneland model accomplishes this by dividing the Monte Carlo (MCMC) runs, with four chains per run, sampling every study area using a Voronoi tessellation and using genetic and 1000 generations. We used TRACER version 1.7.1 (Rambaut et al., geographic distance to calculate the probability of individuals 2018) to assess convergence, accepting once average standard de- originating from the same population (Guillot et al., 2005b). As all viation in split frequencies (ASDSF) reached 0.01 and Effective sequenced birds came from Central and Southeast Sulawesi and are Sample Size (ESS) of model parameters reached 200. These currently grouped as the P. c. rufofuscum subspecies, the current thresholds had been passed by 2 million generations. We exported taxonomy of the species would be supported if this analysis found a 50% majority rule consensus tree from MrBayes and merged it only one cluster. By the same token, multiple clusters would indi- with the Maximum Likelihood tree using the merge_tree function cate more genetic populations than captured by current taxonomy. in the ggtree R package (Yu et al., 2016). This merged tree is shown in Fig. 3, with the outgroup collapsed and with Bayesian probabil- 2.4. Acoustic analysis ities and ML bootstrap values displayed. A version with all outgroup taxa displayed is available in Fig. S2. Comparative analysis of bird song has come to play a central role To check for potential cryptic species within our babblers, we in species delimitation and integrative taxonomy, as differences in carried out distance-based molecular species delimitation using song have been shown to lead to reproductive isolation and Automatic Barcode Gap Discovery (ABGD) (Puillandre et al., 2012). speciation (Isler et al., 1998; O'Reilly et al., 2018). The Sulawesi This analysis uses pairwise genetic distances to group sequences babbler often sings in duet, with the “main song” produced by one into “species” so that genetic distances within these “species” are individual (presumed to be the male of a pair) answered with a smaller than those than between them. It takes a range of prior distinct vocalisation from the presumed female (Billerman et al.,

Fig. 2. Haplotype Network of concatenated babbler ND2/ND3 sequences. Each coloured circle represents a haplotype, sized to represent the number of corresponding samples and coloured to represent our proposed delineation of populations. Each bar across the interconnecting lines represents one mutation. The small, unﬁlled white nodes represent hypothetical ancestral states.

317 F. O Marcaigh, D.J. Kelly, D.P. O'Connell et al. Zoologischer Anzeiger 293 (2021) 314e325

Fig. 3. Consensus phylogenetic tree of babbler haplotypes, incorporating both Bayesian and Maximum Likelihood (ML) analyses. Nodes are labelled with Bayesian probabilities (above, in blue) and ML bootstrap supports (below, in red). The tree shows the outgroup taxa collapsed into one tip, a full list of the outgroup taxa used in the analyses can be found in Table S3 while Fig. S2 shows them included in the tree.

Fig. 4. a) Plot of the Geneland MCMC estimating the number of clusters, i.e. populations at HWE in the babblers of Central and Southeast Sulawesi. b) Histogram summarising the MCMC results and ﬁnding four clusters to be the most likely result. c) Map of babbler concatenated sequences with the clusters they were assigned to by the Geneland analysis.

2020). We analysed the main, “male” song, and not the answering our haplotype network and phylogenetic tree had suggested song of the presumed female, as there were more uninterrupted certain patterns of genetic clustering in the purported rufofuscum song bursts available and male song is more likely to be relevant to subspecies, we were able to test whether this was reflected by song speciation, as mate choice by females based on male song may differences while also extending the comparison to two other reinforce reproductive isolation if song diverges between pop- recognised babbler subspecies (celebense from North Sulawesi and ulations (Catchpole, 1987). We used Raven Pro version 1.6 (Center togianense from Togian). for Conservation Bioacoustics, 2019) to create spectrograms from A random forest is a supervised classification algorithm which the babbler recordings and measured these using on-screen cursors uses a series of decision trees to partition the dataset. A random to collect data. Our acoustic dataset consisted of standard spectral subset of input variables are used to create bootstrapped subsets of and temporal song traits: peak frequency, duration, minimum fre- training data to combine into a final model, splitting the data in a quency, maximum frequency, bandwidth, number of notes, and way that is unbiased and robust (Breiman, 2001). Random forests pace (Tobias et al., 2010; O'Connell et al., 2019b). To aid in visual- have been used to diagnose samples by their origin in other fields, isation, contrast and brightness were set to an equal value and the such as geology (Dornan et al., 2020) and botany (Finch et al., 2017), “Jet” colormap was selected; all other settings were left at their and are increasingly used in similar fashion in species delimitation defaults (Ng et al., 2016). To account for intra-individual variation, and population genetics (Derkarabetian et al., 2019; Smith & intra-individual means were calculated from a minimum of two Carstens, 2020). Diagnosability is key to defining taxa, and is independent bursts of song (average of 8.1 songs, range 2e26) (Ng emphasised in particular by the Phylogenetic Species Concept et al., 2016). These means served as our sample points (Supple- (Cracraft, 1983; Archer et al., 2017). We used the “training set” mentary File 2). setting for WEKA's random forest. A subset of the babbler data was To give an impression of whether babbler populations could be used to train the algorithm, by dividing these training examples by distinguished by song, we used the “random forest” algorithm in subspecies and additional divisions suggested by the genetic and WEKA version 3.8.4 (Frank et al., 2016). This allowed us to look for phylogeographic analyses. In keeping with the current taxonomy, clusters independently in our different datasets. As Geneland and we labelled birds from North Sulawesi as celebense, those from

318 F. O Marcaigh, D.J. Kelly, D.P. O'Connell et al. Zoologischer Anzeiger 293 (2021) 314e325

Togian togianense, and those from Central Sulawesi rufofuscum. similar comparisons with female morphology and the effect sizes Based on genetic divisions we grouped together the birds from were such that the score would have been the same. Southeast Sulawesi and Buton as the “southeast group” and gave separate labels to those from Kabaena. We did not have songs from Wawonii or Muna (Table 1). The algorithm then attempted to 3. Results classify the rest of the data based on differences in the training set. For comparison, we also ran Random Forest analyses which tried to 3.1. Genetics and phylogeography group the birds by island, i.e. with Buton, Southeast Sulawesi, and Muna treated separately. Genetic analysis provided evidence of population structure To confirm the pattern suggested by the Random Forest analysis, within the ostensible P. c. rufofuscum subspecies of Central and we used R to carry out a multivariate ANOVA (or MANOVA) on the Southeast Sulawesi. Southeast Sulawesi shared concatenated ND2- acoustic data to see if differences between songs of populations ND3 haplotypes with Muna and Buton, but not with Wawonii, were statistically significant. This MANOVA tested the difference Kabaena, or Central Sulawesi (Fig. 2). For that reason, in further between the North Sulawesi, Togian, Central Sulawesi, “southeast analyses we grouped Southeast Sulawesi, Buton, and Muna as the group” (Southeast Sulawesi and Buton) and Kabaena populations in “southeast group”, separate to Central Sulawesi. The Central Sula- peak frequency, duration, minimum frequency, high frequency, wesi haplotype (HapRU01) was between 1.7% and 1.9% different bandwidth, number of notes, and pace. We visualised the acoustic from the southeast group haplotypes, while it was 2.4% different data using box plots (Fig. 5). from the Wawonii haplotype (hapWA01) and 1.9e2.2% different from the Kabaena haplotypes (Table S4). Babblers from Kabaena and the southeast group are distinct, monophyletic clusters on the 2.5. Morphological analyses phylogenetic tree (Fig. 3), each diverging from a common ancestor. The Geneland MCMC analysis found best support for a 4-cluster Only adult birds measured by NMM were included in morpho- model (Fig. 4), followed by a 3-cluster model. This indicates that metric analyses, and male and female babblers were analysed there is more population structure in these babbler populations separately as males are considerably larger (O Marcaigh et al., than is recognised by the current taxonomy, where all of these birds 2020). The morphometric analyses thus included only birds from are considered to belong to one subspecies (P. c. rufofuscum). the Kabaena (n ¼ 18) and “southeast group” (n ¼ 50) populations, Furthermore, its proposed populations line up with the divisions where the southeast group comprised Southeast Sulawesi, Muna, suggested by the phylogenetic tree, with the Kabaena, Wawonii, and Buton. We selected wing length, bill length, and skull length for southeast group, and Central Sulawesi clusters all apparent the morphological analyses as they are independent variables that (Fig. 4c). correspond to different aspects of the birds’ ecology (O Marcaigh The ABGD analysis found the most support for three species in et al., 2020). As with the acoustic data, we first carried out a Pellorneum (P. rostratum, P. bicolor, and P. celebense), with five random forest classification to see if these populations could be different barcode gaps between 0.8% and 6% difference in concat- distinguished based on these three traits, then a MANOVA to test enated ND2-ND3 producing this grouping. At the lowest prior whether the differences were statistically significant. We con- intraspecific divergences, however, the analysis calculated the structed box plots to visualise the morphological data (Fig. 6). barcode gap distance to be 0.1% and with this it created six groups: P. rostratum, P. bicolor, P. celebense from Central Sulawesi, P. celebense from Wawonii, P. celebense from Kabaena, and 2.6. Tobias scoring P. celebense from the southeast group. We subjected our putative babbler populations to the quantitative scoring criteria outlined by Tobias et al. (2010), where populations with a “score” of 7 or more are seen as deserving of species 3.2. Acoustic results status. This score is a combination of differences in morphology, vocalisations, plumage, ecology or behaviour, and geography. The The Random Forest analysis on the seven acoustic traits had a geography category awards points for situations of sympatry and 100% success rate in classifying birds according to our population hybrid zones, and so does not apply to island populations. We had divisions of Togian, North Sulawesi, Central Sulawesi, the “south- no data on ecology or behaviour, and our photographs of live birds east group” (Southeast Sulawesi and Buton), and Kabaena. The in the field (Fig. S1) could not demonstrate plumage differences due Kappa statistic, F-Measure and ROC Area for each division was to varying light conditions, so we were unable to score those areas. equal to 1, indicating an optimal classifying model. MANOVA on the Therefore we calculated a partial Tobias score for the Kabaena seven traits agreed, finding a statistically significant difference population on acoustics and morphology, using the “effsize” between the same population divisions (p < 0.001). The full output package in R (Torchiano, 2020) to calculate Cohen's d. Effect sizes from this MANOVA is in the Supplementary Information (Table S5). are more suitable than p-values for informing taxonomic judge- The Random Forest acoustic analysis which treated each island ments as they are less correlated with sample size (Tobias et al., individually was much less successful, as it classified 31 birds 2010). We used male morphology for our Tobias score because incorrectly (48%). The Kappa statistic was 0.37, indicating a sub- the acoustic score was based on the male song. We also made optimal model.

Table 1 Number of individual babblers from each area available for each analysis. Mainland Sulawesi areas are bolded, offshore islands are italicised.

n Southeast Sulawesi Central Sulawesi North Sulawesi Togian Kabaena Muna Buton Wawonii

DNA 5 1 0 0 8 6 5 1 Morphology 22 0 0 0 18 6 22 0 Song 15 2 12 2 17 0 16 0

319 F. O Marcaigh, D.J. Kelly, D.P. O'Connell et al. Zoologischer Anzeiger 293 (2021) 314e325

Fig. 5. Box plots of acoustic traits of Sulawesi babblers from Central Sulawesi, Kabaena, North Sulawesi, the Southeast Group (including mainland Southeast Sulawesi and Buton), and Togian.

Fig. 6. Box plots of morphological traits of Sulawesi babblers from Kabaena and the Southeast Group (including mainland Southeast Sulawesi, Buton, and Muna). The top row of box plots represent the male babblers, the bottom row the females.

320 F. O Marcaigh, D.J. Kelly, D.P. O'Connell et al. Zoologischer Anzeiger 293 (2021) 314e325

3.3. Morphological results Pellorneum celebense rufofuscum is actually comprised of four independently evolving lineages. Once juvenile birds were excluded from the morphological The babblers of Southeast Sulawesi were formerly recognised as dataset, this left only the southeast group and the Kabaena popu- a distinct subspecies, P. c. improbatum, originally described by lation. Even so, the Random Forest analysis categorised each Stresemann (1938) from a type specimen from Lalolai in Southeast dataset with 100% accuracy and with Kappa statistics, F-Measures Sulawesi (latitude 4.05, longitude 121.88). Though this taxon was and ROC Areas equal to 1. The MANOVA on male babblers from the abolished by White & Bruce (1986) and Gill et al. (2021), our ana- southeast group and from Kabaena found that they were statisti- lyses have reaffirmed that the babblers of the southeast group cally significantly different in wing length, bill length and skull (Southeast Sulawesi, Buton, and Muna), do in fact represent an length (p < 0.05), as did the MANOVA on female babblers from the evolutionarily distinct population. The Southeast population has same two populations, using the same traits (p < 0.05). The full diverged from Central Sulawesi both acoustically and morpholog- output from these MANOVAs are in the Supplementary Information ically, mirroring the patterns of endemism seen in trees, monkeys, (male babblers in Table S6, females in Table S7). When treating each toads, and hoofed mammals (Evans et al., 2003; Frantz et al., 2018; island separately, the accuracy of morphological Random Forest Trethowan et al., 2020). This supports the hypothesis that these analysis on males declined to 35%, while accuracy on females was understorey birds are as disinclined to disperse as are non-volant reduced to 40%. organisms, with corresponding impacts on evolutionary trajec- tories. This is in keeping with findings from other understorey 3.4. Tobias scoring babbler species, which have also diverged genetically in areas that were recently connected by land bridges (Cros et al., 2020). We applied the Tobias species delimitation system (Tobias et al., The distances involved would appear to be too short to fully 2010) to our division between the Kabaena population and the explain the divergence seen. Kabaena is the most distant of the southeast group (consisting of Southeast Sulawesi, Muna, and land-bridge islands at around 18 km from the Sulawesi mainland Buton), because this division had been supported by all of our (Robinson-Dean et al., 2002), still a relatively short distance in previous analyses. This system allows use of one spectral acoustic terms of speciation. Wawonii is only 7 km from Sulawesi, compa- character and one temporal in distinguishing between bird pop- rable to Buton which is 6 km from the mainland at its closest point. ulations. Peak frequency was the most divergent spectral character Muna is separated from Buton by only 0.6 km. Habitats on Kabaena between Kabaena and the southeast group, with a Cohen's d of 0.36. and Wawonii may be isolated by geology more than distance, as Pace was the most divergent temporal character, with a Cohen's d of both islands are dominated by a distinct ultramafic geology that 1.77. These constitute “minor” differences under the Tobias system, produces soils poor in nutrients and rich in phytotoxic minerals scoring 1 point each. (Galey et al., 2017). Such soils present distinct selection pressures Two morphological traits may be included in calculating a for organisms and are noted for very high levels of plant endemism Tobias score: that showing the largest increase, and that showing (Anacker, 2014), which would in turn present a distinct evolu- the largest decrease. When comparing birds from Kabaena to those tionary environment for animals including babblers. Populations of from the southeast group, these were wing length (Cohen's d of other taxa on Kabaena and Wawonii have been noted for their 1.28) and bill length (Cohen's d of 0.25). The system ranks these distinctness from neighbouring islands, showing the potential for effect sizes as “minor” and awards 1 point for each of them. Added evolutionary and ecological divergence despite the short time scale. together, these give a Tobias score of 4 for the Kabaena population Tweedley et al. (2013) observed a pronounced difference between when compared to babblers from Southeast Sulawesi, Buton, and the composition of the freshwater fish faunas of Kabaena and Muna. Buton. Trethowan et al. (2020) found the tree communities of Wawonii to comprise different species than those of Central 4. Discussion Sulawesi. Within birds, the patterns of evolutionary divergence in the Wallace (1887) set the priorities of biogeographers for centuries region can be linked to life history. Zosterops white-eyes (Vigors & when he wrote that “The continental islands, still attached as they Horsfield, 1826) are famous for their dispersal abilities, though are to the base of the mainland, are to all intents and purposes a more isolated populations are known to develop behavioural portion of the continent, as well in structure as in the forms of flightlessness (Moyle et al., 2009). Zosterops populations on animal and vegetable life which they afford. It is in the oceanic Kabaena and Wawonii show no sign of divergence (O'Connell et al., islands that we should meet with limited and peculiar types.” 2019b). The olive-backed sunbird Cinnyris jugularis (Linnaeus, 1766) Oceanic islands continue to receive the most attention in speciation and grey-sided flowerpecker Dicaeum celebicum (Müller, 1843) studies today (Tobias et al., 2020). While continental land-bridge both inhabit marginal habitats such as forest edges, scrubland and islands indeed harbour fewer endemic species, their populations mangroves (Billerman et al., 2020), a trait that makes a bird more may yet represent unique components of the species, important for likely to cross habitat gaps and open spaces (Burney & Brumfield, both current biodiversity and future evolution. The fact that land- 2009). They too share populations across Kabaena and Sulawesi bridge islands are numerous and tend to be richer in species (Kelly, 2014; Kelly et al., 2014; O'Connell et al., 2019c). The red- makes their populations more interesting still (Meiri, 2017; Tobias backed thrush Geokichla erythronota (Sclater, 1859) presents a et al., 2020). It is apparent that babblers on the land-bridge islands notable contrast to these species, having diverged strongly enough of Kabaena and Wawonii have diverged from mainland populations on Kabaena to produce an endemic subspecies G. e. kabaena in the brief time since these landmasses became physically (Robinson-Dean et al., 2002). Like the Sulawesi babbler, the rusty- disconnected. Previous work has shown that babblers on the land- backed thrush is a bird of the forest understorey, and the two bridge islands of Sulawesi exhibit stronger sexual dimorphism than species have even been observed foraging together (Billerman those on the mainland (O Marcaigh et al., 2020). This study adds et al., 2020). that babblers on Kabaena are distinct in acoustics, morphology, and Evolutionary divergence within the babblers of Sulawesi and its mtDNA, and that the Wawonii population is strongly divergent in land-bridge islands can be compared to that between more mtDNA. Combined with the division we found between Southeast geographically isolated lineages. This study found Moyle et al.’s Sulawesi and Central Sulawesi, this indicates that the subspecies (2012) P. rostratum sequence from Borneo to be 7% different from

321 F. O Marcaigh, D.J. Kelly, D.P. O'Connell et al. Zoologischer Anzeiger 293 (2021) 314e325 all Pellorneum celebense haplotypes. Borneo and Sulawesi have species have ever been collected there, in fact the island was almost been separated for the duration of their existence by the Makassar unknown ornithologically until recently (O'Connell et al., 2017). Strait, part of the permanent barrier of deep water known to bio- Lack of museum material should not delay conservation (O'Connell geographers as Wallace's Line (Wallace, 1880; Tweedley et al., et al., 2020). We therefore suggest the provisional name of P. c. 2013). This 7% difference in concatenated ND2-ND3 is thus the kabaena for the Kabaena babblers and recommend that they be result of millions of years of evolution, while the populations of subjected to formal description. The Wawonii population shows Sulawesi and its land-bridge islands have developed a p-distance strong divergence in mtDNA, but as this is based on one sample and around 1/3 of this in only 12,000 years or so (Table S4). This illus- lacks acoustic or morphological data, we recommend that it be trates the remarkable speed with which speciation can act on studied in more detail before a taxonomic judgement is made. dispersal-limited species, as well as the impact of genetic drift on Separate to any taxonomic revision, characterising the diver- these relatively small island populations. gence of populations on Wallacean islands is key to our under- Our work highlights the importance of units below the species standing of how speciation creates the biodiversity of this global level to the evolutionary potential of the Sulawesi babble. This joins hotspot. The concept of the Evolutionarily Significant Unit, or ESU, a long-standing debate in evolutionary biology, where approaches aims to sidestep taxonomic debate by targeting conservation at to species and subspecies often generate controversy. Earlier nat- lineages that are evolving independently, regardless of how these uralists named many subspecies based on typological traits and are assigned to taxa (Ryder, 1986; Moritz, 1994; Coates et al., 2018; came in for some contemporary criticism (Wilson & Brown, 1953). Neal et al., 2018). We believe that four independent ESUs are pre- While modern methods overturned many of their designations, sent in the populations currently assigned to this one babbler they reaffirmed others: Hartert (1903), for example, named two subspecies, one each from Central Sulawesi, Southeast Sulawesi, endemic species from the Wakatobi islands which were later Kabaena, and Wawonii. demoted to subspecies (White & Bruce, 1986), before being These populations of the supposed rufofuscum subspecies have confirmed as reproductively isolated species a century later (Kelly undergone divergent evolution, despite being physically capable of et al., 2014; O'Connell et al., 2019b). The higher levels of the maintaining gene flow between the land-bridge islands and the Linnaean hierarchy were prioritised early in the molecular age, but mainland. As Mayr (1969) noted, “Most tropical birds are highly a broad consensus persisted that the subspecies concept is useful in sedentary and respect water barriers to a high degree”. This also naming distinct populations with geographical boundaries be- applies to the division between Central Sulawesi and Southeast tween them, using multiple lines of evidence (Wiens et al., 1982). Sulawesi, which were separated by water barriers for much of their Phillimore & Owens (2006) found that island subspecies are the geological history. Taxonomists should be wary of lumping the taxa most likely to reflect evolution accurately and suggested that sub- of these two “areas of endemism” (Evans et al., 2003) together. Just species can aid conservation in the tropics. We have sought to as our understanding of a species and its evolution is improved by follow these recommendations by analysing geographically analysing its constituent populations, there is an urgent need to delimited populations on tropical islands using multiple lines of consider which areas within the threatened biodiversity hotspots evidence. are most important in generating this biodiversity through speci- Modern ornithologists delimit species with an integrative ation. Sometimes these areas are imperilled by the very traits that approach based on comparison to recognised species and ulti- make them evolutionarily significant: the ultramafic nickel de- mately derived from the Biological Species Concept (Tobias et al., posits of Kabaena and Wawonii also make them attractive to large- 2010). This integrative approach is particularly important when scale mining (Morse, 2019b, a). Time is therefore running out to genetic data are absent or, as here, based on relatively low genetic build a full picture of the biodiversity of these islands and their sample sizes. Based on two of the five scoring criteria, the Kabaena evolutionary dynamics. population attains a Tobias score of 4, three points short of species status. Despite the lack of data on plumage, ecology, and behaviour, Funding this partial score was more than half of that required to identify a distinct species, strongly suggesting that the Kabaena birds and the This project was funded by the Irish Research Council under southeast group represent separate populations. The Wawonii application (GOIPG/2017/1618). population cannot be scored as it lacks acoustic and morphological data. Note that this system does not incorporate genetic data. We Availability of data and material thus propose the distinct babbler populations as subspecies and not as independent species. Our ABGD analysis supports this conclu- All new ND2 and ND3 DNA sequences from this study have been sion, as it separated these four P. celebense populations from one submitted to GenBank under accession numbers MW387438 - another using a small barcode gap of 0.1% difference in concate- MW387487. Recordings of vocalisations will be uploaded to xeno- nated ND2-ND3, appropriate for splits between subspecies. When canto. The acoustic and morphological data have been made the barcode gap was between 0.8% and 6%, similar to that available in Supplementary Files 2 and 3, respectively. betweenND2 sequences of other bird species (e.g. Pellegrino et al., 2017; O'Connell et al., 2019b), ABGD grouped P. celebense together as a single species. Fuller Tobias assessments, incorporating Code availability plumage, ecology, and behaviour, might lead to other conclusions. We thus recommend that the subspecies Pellorneum celebense The R code used in our analyses is available in Supplementary improbatum be reinstated for babblers from Southeast Sulawesi, File 1. Buton, and Muna, as these are genetically and acoustically divergent from the P. c. rufofuscum population of Central Sulawesi. We Authors’ contributions propose that babblers from Kabaena be named as a new subspecies for their genetic, acoustic, and morphological divergence from FOM, DJK, and NMM conceived this study and led the writing of Southeast Sulawesi, Buton, and Muna. This would require collection the manuscript. DJK, NMM, DOC, FOM, AK, and KA carried out the of a voucher specimen from Kabaena, as historical collectors did not fieldwork. FOM, NL, DOC, DD, and AC carried out the labwork. FOM visit the island (White & Bruce, 1986) and as a result very few and DD processed the song recordings and DNA sequences. FOM

322 F. O Marcaigh, D.J. Kelly, D.P. O'Connell et al. Zoologischer Anzeiger 293 (2021) 314e325 analysed the data, with input from DOC on phylogenetics. All au- between mitochondrial and nuclear markers. Mitochondrial DNA 4, e thors contributed to revising and improving the manuscript. 2566 2569. https://doi.org/10.1080/23802359.2019.1637286. Carstens, B.C., Pelletier, T.A., Reid, N.M., Satler, J.D., 2013. How to fail at species delimitation. Mol. Ecol. 22, 4369e4383. https://doi.org/10.1111/mec.12413. Ethics approval Catchpole, C.K., 1987. Bird song, sexual selection and female choice. Trends Ecol. Evol. 2, 94e97. https://doi.org/10.1016/0169-5347(87)90165-0. Center for Conservation Bioacoustics, 2019. Raven Pro: Interactive Sound Analysis Kementerian Riset Teknologi Dan Pendidikan Tinggi (RIS- Software. Version 1, 6 ed. The Cornell Lab of Ornithology, Ithaca, NY. TEKDIKTI) provided the necessary permits and approvals, under Chesser, R.T., 1999. Molecular systematics of the rhinocryptid genus Pteroptochos. permit numbers 0143/SlP/FRP/SM/Vll/2010, 278/SIP/FRP/SM/Vll/ Condor 101, 439e446. http://www.jstor.org/stable/1370012. Cibois, A., 2003. Mitochondrial DNA phylogeny of babblers (Timaliidae). Auk 120, 2012, 279/SIP/FRP/SM/VIII/2012, 174/SIP/FRP/E5/Dit. KI/V/2016, 35e54. https://doi.org/10.1093/auk/120.1.35. 159/SIP/FRP/E5/Fit.KIVII/2017, and 160/ SIP/FRP/E5/Fit.KIVII/2017. Cibois, A., Kalyakin, M.V., Lian-Xian, H., Pasquet, E., 2002. Molecular phylogenetics of babblers (Timaliidae): revaluation of the genera Yuhina and Stachyris. J. Avian Biol. 33, 380e390. https://doi.org/10.1034/j.1600-048X.2002.02882.x. Declaration of competing interest Coates, D.J., Byrne, M., Moritz, C., 2018. Genetic diversity and conservation units: dealing with the species-population continuum in the Age of Genomics. Front. The authors declare that they have no known competing Ecol. Evol. 6 https://doi.org/10.3389/fevo.2018.00165. fi Cracraft, J., 1983. Species concepts and speciation analysis. In: Johnson, R.F. (Ed.), nancial interests or personal relationships that could have Current Ornithology. Springer, New York, pp. 159e187. appeared to influence the work reported in this paper. Cros, E., Chattopadhyay, B., Garg, K.M., Ng, N.S.R., Tomassi, S., Benedick, S., Edwards, D.P., Rheindt, F.E., 2020. Quaternary land bridges have not been uni- fl e Acknowledgements versal conduits of gene ow. Mol. Ecol. 29, 2692 2706. https://doi.org/10.1111/ mec.15509. Darwin, C., 1859. On the Origin of Species by Means of Natural Selection, or the The Irish Research Council funded this research and Kemen- Preservation of Favoured Races in the Struggle for Life. John Murray, London. terian Riset Teknologi Dan Pendidikan Tinggi (RISTEKDIKTI) pro- Deignan, H., 1964. Family muscicapidae, subfamily timaliinae. In: Mayr, E., Paynter, R., Jnr (Eds.), Checklist of Birds of the World. Harvard University Press, vided the necessary permits and approvals (see Declarations for Cambridge, pp. 240e427. https://doi.org/10.5962/bhl.title.14581. details). In addition to these bodies, we thank Operation Wallacea Derkarabetian, S., Castillo, S., Koo, P.K., Ovchinnikov, S., Hedin, M., 2019. for the logistical support they provided to our many field expedi- A demonstration of unsupervised machine learning in species delimitation. fi Mol. Phylogenet. Evol. 139, 106562 https://doi.org/10.1016/ tions, and everyone who enabled our eldwork via access to land j.ympev.2019.106562. and gracious hospitality. Finally, we thank two anonymous re- Diamond, J.M., 1981. Flightlessness and fear of flying in island species. Nature 293, viewers for improving the manuscript with their helpful 507e508. Diamond, J.M., 1985. Taxonomy: how many unknown species are yet to be comments. discovered? Nature 315, 538e539. https://doi.org/10.1038/315538a0. Dobzhansky, T., 1937. Genetics and the Origin of Species. Columbia University Press, Appendix A. Supplementary data New York. Dobzhansky, T., 1940. Speciation as a stage in evolutionary divergence. Am. Nat. 74, 312e321. https://doi.org/10.1086/280899. Supplementary data to this article can be found online at Dobzhansky, T., Pavlovsky, O., 1967. Experiments on the incipient species of the https://doi.org/10.1016/j.jcz.2021.07.006. Drosophila paulistorum complex. Genetics 55, 141e156. Dornan, T., O'Sullivan, G., O'Riain, N., Stueeken, E., Goodhue, R., 2020. The application of machine learning methods to aggregate geochemistry predicts quarry References source location: an example from Ireland. Comput. Geosci. 140, 104495 https:// doi.org/10.1016/j.cageo.2020.104495. Ali, J.R., 2017. Islands as biological substrates: classification of the biological Esper, E.J.C., 1781. De Varietatibus Specierum in Naturae Productis. Erlangen. assemblage components and the physical island types. J. Biogeogr. 44, 984e994. Evans, B.J., Supriatna, J., Andayani, N., Setiadi, M.I., Cannatella, D.C., Melnick, D.J., Ali, J.R., 2018. Islands as biological substrates: continental. J. Biogeogr. 45, 2003. Monkeys and toads define areas of endemism on Sulawesi. Evolution 57, 1003e1018. https://doi.org/10.1111/jbi.13186. 1436e1443. https://doi.org/10.1111/j.0014-3820.2003.tb00350.x. Anacker, B.L., 2014. The nature of serpentine endemism. Am. J. Bot. 101, 219e224. Everson, K.M., Hildebrandt, K.B.P., Goodman, S.M., Olson, L.E., 2018. Caught in the https://doi.org/10.3732/ajb.1300349. act: incipient speciation across a latitudinal gradient in a semifossorial mammal Andersen, M.J., Shult, H.T., Cibois, A., Thibault, J.C., Filardi, C.E., Moyle, R.G., 2015. from Madagascar, the mole tenrec Oryzorictes hova (Tenrecidae). Mol. Phylo- Rapid diversification and secondary sympatry in australo-pacific kingfishers genet. Evol. 126, 74e84. https://doi.org/10.1016/j.ympev.2018.02.024. (aves: alcedinidae: todiramphus). Royal Soc. Open Sci. 2, 140375 https://doi.org/ Finch, K., Espinoza, E., Jones, F.A., Cronn, R., 2017. Source identification of western 10.1098/rsos.140375. Oregon Douglas-fir wood cores using mass spectrometry and random forest Archer, F.I., Martien, K.K., Taylor, B.L., 2017. Diagnosability of mtDNA with Random classification. Appl. Plant Sci. 5, 1600158 https://doi.org/10.3732/apps.1600158. Forests: using sequence data to delimit subspecies. Mar. Mamm. Sci. 33, Flantua, S.G.A., Payne, D., Borregaard, M.K., Beierkuhnlein, C., Steinbauer, M.J., 101e131. https://doi.org/10.1111/mms.12414. Dullinger, S., et al., 2020. Snapshot isolation and isolation history challenge the Bickford, D., Lohman, D.J., Sodhi, N.S., Ng, P.K., Meier, R., Winker, K., et al., 2007. analogy between mountains and islands used to understand endemism. Global Cryptic species as a window on diversity and conservation. Trends Ecol. Evol. Ecol. Biogeogr. 29, 1651e1673. https://doi.org/10.1111/geb.13155. 22, 148e155. https://doi.org/10.1016/j.tree.2006.11.004. Frank, E., Hall, M.A., Witten, I.H., 2016. The WEKA workbench. In: Witten, I.H., Billerman, S.M., Keeney, B.K., Rodewald, P.G., Schulenberg, T.S., 2020. Birds of the Frank, E., Hall, M.A., Pal, C.J. (Eds.), Data Mining: Practical Machine Learning World. Cornell Laboratory of Ornithology. https://birdsoftheworld.org/bow/ Tools and Techniques, fourth ed. Morgan Kaufmann, Burlington, MA, pp. 1e128 home. (Accessed 10 June 2020). https://www.cs.waikato.ac.nz/ml/weka/citing.html. Blyth, E., 1842. Report from the curator. J. Asiatic Soc. Bengal 11, 788e809. https:// Frantz, L.A.F., Rudzinski, A., Nugraha, A.M.S., Evin, A., Burton, J., Hulme-Beaman, A., www.biodiversitylibrary.org/page/40005158. et al., 2018. Synchronous diversification of Sulawesi's iconic artiodactyls driven Breiman, L., 2001. Random forests. Mach. Learn. 45, 5e32. https://doi.org/10.1023/ by recent geological events. Proc. R. Soc. B 285. https://doi.org/10.1098/ A:1010933404324. rspb.2017.2566. Brelsford, A., Irwin, D.E., 2009. Incipient speciation despite little assortative mating: Galey, M.L., van der Ent, A., Iqbal, M.C.M., Rajakaruna, N., 2017. Ultramafic geo- the yellow-rumped warbler hybrid zone. Evolution 63, 3050e3060. https:// ecology of South and Southeast asia. Bot. Stud. 58, 18. https://doi.org/10.1186/ doi.org/10.1111/j.1558-5646.2009.00777.x. s40529-017-0167-9. Burney, C.W., Brumfield, R.T., 2009. Ecology predicts levels of genetic differentiation Gaston, K.J., Blackburn, T.M., 1994. Are newly described bird species small-bodied? in neotropical birds. Am. Nat. 174, 358e368. https://doi.org/10.1086/603613. Biodivers. Lett. 2, 16e20. https://doi.org/10.2307/2999696. Cai, T., Cibois, A., Alstrom,€ P., Moyle, R.G., Kennedy, J.D., Shao, S., et al., 2019. Near- Gelang, M., Cibois, A., Pasquet, E., Olsson, U., Alstrom,€ P., Ericson, P.G.P., 2009. complete phylogeny and taxonomic revision of the world's babblers (Aves: Phylogeny of babblers (Aves, Passeriformes): major lineages, family limits and passeriformes). Mol. Phylogenet. Evol. 130, 346e356. https://doi.org/10.1016/ classification. Zool. Scripta 38, 225e236. https://doi.org/10.1111/j.1463- j.ympev.2018.10.010. 6409.2008.00374.x. Cai, T., Shao, S., Kennedy, J.D., Alstrom,€ P., Moyle, R.G., Qu, Y., et al., 2020. The role of Gill, F., Donsker, D.B., Rasmussen, P.C., 2021. IOC World Bird List (v11.1). https:// evolutionary time, diversification rates and dispersal in determining the global www.worldbirdnames.org/. (Accessed 7 May 2021). diversity of a large radiation of passerine birds. J. Biogeogr. 47, 1612e1625. Guillot, G., Mortier, F., Estoup, A., 2005a. Geneland: a computer package for land- https://doi.org/10.1111/jbi.13823. scape genetics. Mol. Ecol. Notes 5, 712e715. https://doi.org/10.1111/j.1471- Campillo, L.C., Burns, K.J., Moyle, R.G., Manthey, J.D., 2019. Mitochondrial genomes 8286.2005.01031.x. of the bird genus Piranga: rates of sequence evolution, and discordance

323 F. O Marcaigh, D.J. Kelly, D.P. O'Connell et al. Zoologischer Anzeiger 293 (2021) 314e325

Guillot, G., Estoup, A., Mortier, F., Cosson, J.F., 2005b. A spatial statistical model for Meiri, S., 2017. Oceanic island biogeography: nomothetic science of the anecdotal. landscape genetics. Genetics 170, 1261e1280. https://doi.org/10.1534/ Front. Biogeogr. 9, e32081 https://doi.org/10.21425/F59132081. genetics.104.033803. Moritz, C., 1994. Defining 'evolutionarily significant units' for conservation. Trends Hall, T.A., 1999. BioEdit: a user-friendly biological sequence alignment editor and Ecol. Evol. 9, 373e375. https://doi.org/10.1016/0169-5347(94)90057-4. analysis program for Windows 95/98/NT. Nucleic Acids Symp. Ser. 41, 95e98. Morse, I., 2019a. In: Indonesia, a Tourism Village Holds off a Nickel Mine d for Now. Hall, R., 2013. The palaeogeography of Sundaland and Wallacea since the late Mongabay. https://news.mongabay.com/2019/12/in-indonesia-a-tourism- jurassic. J. Limnol. 72, 1e17. https://doi.org/10.4081/jlimnol.2013.s2.e1. village-holds-off-a-nickel-mine-for-now/. (Accessed 30 October 2020). Harris, R.J., Reed, J.M., 2002. Behavioral barriers to non-migratory movements of Morse, I., 2019b. Indonesian Villagers Fighting Planned Mine Garner National birds. Ann. Zool. Fenn. 39, 275e290. Support. Mongabay. https://news.mongabay.com/2019/10/indonesian- Hartert, E., 1903. by Mr Heinrich Kühn. Novit. Zool. On the Birds Collected on the villagers-fighting-planned-mine-garner-national-support/. (Accessed 30 Tukang-Besi Islands and Buton, South-East of Celebes, vol. 10, pp. 18e38. October 2020). Harvey, M.G., Seeholzer, G.F., Smith, B.T., Rabosky, D.L., Cuervo, A.M., Brumfield, R.T., Moyle, R.G., Filardi, C.E., Smith, C.E., Diamond, J., 2009. Explosive Pleistocene 2017. Positive association between population genetic differentiation and diversification and hemispheric expansion of a "great speciator. Proc. Natl. speciation rates in New World birds. Proc. Natl. Acad. Sci. U.S.A. 114, Acad. Sci. U. S. A. 106, 1863e1868. https://www.ncbi.nlm.nih.gov/pubmed/ 6328e6333. https://doi.org/10.1073/pnas.1617397114. 19181851. Hebert, P.D., Stoeckle, M.Y., Zemlak, T.S., Francis, C.M., 2004. Identification of birds Moyle, R.G., Andersen, M.J., Oliveros, C.H., Steinheimer, F.D., Reddy, S., 2012. Phy- through DNA barcodes. PLoS Biol. 2, e312. https://doi.org/10.1371/ logeny and biogeography of the core babblers (Aves: Timaliidae). Syst. Biol. 61, journal.pbio.0020312. 631e651. https://doi.org/10.1093/sysbio/sys027. Hebert, P.D., Ratnasingham, S., Zakharov, E.V., Telfer, A.C., Levesque-Beaudin, V., Müller, S., 1843. Land-en volkenkunde. In: Temminck, C.J. (Ed.), Verhandelingen Milton, M.A., et al., 2016. Counting animal species with DNA barcodes: Canadian over de natuurlijke geschiedenis der Nederlandsche overzeesche bezittingen. insects. Philos. Trans. R. Soc. B 371. https://doi.org/10.1098/rstb.2015.0333. Leiden Leiden. https://doi.org/10.5962/bhl.title.114730. Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference of phylogenetic Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A.B., Kent, J., 2000. trees. Bioinformatics 17, 754e755. https://doi.org/10.1093/bioinformatics/ Biodiversity hotspots for conservation priorities. Nature 403, 853e858. https:// 17.8.754. doi.org/10.1038/35002501. Isler, M.L., Isler, P.R., Whitney, B.M., 1998. Use of vocalizations to establish species Neal, K.M., Johnson, B.B., Shaffer, H.B., 2018. Genetic structure and environmental limits in antbirds (Passeriformes: thamnophilidae). Auk 115, 577e590. https:// niche modeling confirm two evolutionary and conservation units within the doi.org/10.2307/4089407. western spadefoot (Spea hammondii). Conserv. Genet. 19, 937e946. https:// Itescu, Y., Foufopoulos, J., Pafilis, P., Meiri, S., Borregaard, M.K., 2020. The diverse doi.org/10.1007/s10592-018-1066-7. nature of island isolation and its effect on land bridge insular faunas. Global Ng, E.Y.X., Eaton, J.A., Verbelen, P., Hutchinson, R.O., Rheindt, F.E., 2016. Using bio- Ecol. Biogeogr. 29, 262e280. https://doi.org/10.1111/geb.13024. acoustic data to test species limits in an Indo-Pacific island radiation of Mac- Kelly, S.B.A., 2014. Integrative Avian Taxonomy and the Role of Competition in the ropygia cuckoo doves. Biol. J. Linn. Soc. 118, 786e812. https://doi.org/10.1111/ Diversification of Passerine Birds. PhD Thesis (Unpublished), Trinity College bij.12768. Dublin. Nugraha, A.M.S., Hall, R., 2018. Late cenozoic palaeogeography of Sulawesi, Kelly, S.B.A., Kelly, D.J., Cooper, N., Bahrun, A., Analuddin, K., Marples, N.M., 2014. Indonesia. Palaeogeogr. Palaeoclimatol. Palaeoecol. 490, 191e209. https:// Molecular and phenotypic data support the recognition of the Wakatobi doi.org/10.1016/j.palaeo.2017.10.033. Flowerpecker (Dicaeum kuehni) from the unique and understudied Sulawesi O Marcaigh, F., Kelly, D.J., Analuddin, K., Karya, A., Lawless, N., Marples, N.M., 2020. region. PloS One 9, e98694. https://doi.org/10.1371/journal.pone.0098694. Cryptic sexual dimorphism reveals differing selection pressures on continental Keogh, J.S., Scott, I.A., Hayes, C., 2005. Rapid and repeated origin of insular gigantism islands. Biotropica 53, 121e129. https://doi.org/10.1111/btp.12852. and dwarfism in Australian tiger snakes. Evolution 59, 226e233. https:// O'Brien, S.J., Mayr, E., 1991. Bureaucratic mischief: recognizing endangered species doi.org/10.1111/j.0014-3820.2005.tb00909.x. and subspecies. Science 251, 1187e1189. https://doi.org/10.1126/ Kerr, K.C.R., Stoeckle, M.Y., Dove, C.J., Weigt, L.A., Francis, C.M., Hebert, P.D.N., 2007. science.251.4998.1187. Comprehensive DNA barcode coverage of North American birds. Mol. Ecol. O'Connell, D.P., Sealy, S., O Marcaigh, F., Karya, A., Bahrun, A., Analuddin, K., et al., Notes 7, 535e543. https://doi.org/10.1111/j.1471-8286.2007.01670.x. 2017. The avifauna of kabaena island, south-east Sulawesi, Indonesia. Forktail Klein, E.R., Harris, R.B., Fisher, R.N., Reeder, T.W., 2016. Biogeographical history and 40e45. coalescent species delimitation of Pacific island skinks (Squamata: scincidae: O'Connell, D.P., O Marcaigh, F., O'Neill, A., Griffin, R., Karya, A., Analuddin, K., et al., Emoia cyanura species group). J. Biogeogr. 43, 1917e1929. https://doi.org/ 2019a. The avifauna of Muna and Wawonii Island, with additional records from 10.1111/jbi.12772. mainland south-east Sulawesi, Indonesia. Forktail 35, 50e56. Korbie, D.J., Mattick, J.S., 2008. Touchdown PCR for increased specificity and O'Connell, D.P., Kelly, D.J., Lawless, N., O'Brien, K., O Marcaigh, F., Karya, A., et al., sensitivity in PCR amplification. Nat. Protoc. 3, 1452e1456. https://doi.org/ 2019b. A sympatric pair of undescribed white-eye species (Aves: zosteropidae: 10.1038/nprot.2008.133. Zosterops) with different origins. Zool. J. Linn. Soc. 186, 1e24. https://doi.org/ Kumar, S., Stecher, G., Li, M., Knyaz, C., Tamura, K., 2018. MEGA X: molecular 10.1093/zoolinnean/zlz022. evolutionary genetics analysis across computing platforms. Mol. Biol. Evol. 35, O'Connell, D.P., Kelly, D.J., Kelly, S.B., Sealy, S., Karya, A., Analuddin, K., et al., 2019c. 1547e1549. https://doi.org/10.1093/molbev/msy096. Increased sexual dimorphism in dense populations of Olive-backed Sunbirds on Leigh, J.W., Bryant, D., Nakagawa, S., 2015. POPART: full-feature software for small islands: morphological niche contraction in females but not males. Emu haplotype network construction. Methods Ecol. Evol. 6, 1110e1116. https:// 119, 296e307. https://doi.org/10.1080/01584197.2019.1588743. doi.org/10.1111/2041-210X.12410. O'Connell, D.P., Kelly, D.J., Analuddin, K., Karya, A., Marples, N.M., Martin, T.E., 2020. Lesson, R.P., 1839. Oiseaux rares ou nouveaux de la collection du docteur Abeille, de Adapt taxonomy to conservation goals. Science 369, 1172. https://doi.org/ Bordeaux. Revue zoologique 1, 136e139. https://www.biodiversitylibrary.org/ 10.1126/science.abd7717. page/2299342. O'Reilly, C., Analuddin, K., Kelly, D.J., Harte, N., 2018. Measuring vocal difference in Linnaeus, C., 1766. Certhia Jugularis. Systema Naturae Per Regna Tria Naturae, bird population pairs. J. Acoust. Soc. Am. 143, 1658. https://doi.org/10.1121/ Secundum Classes, Ordines, Genera, Species, Cum Characteribus, Differentiis, 1.5027244. Synonymis, Locis. Tomus I. Editio Duodecima, Reformata., Stockholm, p. 185. Pellegrino, I., Cucco, M., Harvey, J.A., Liberatore, F., Pavia, M., Voelker, G., et al., 2017. https://doi.org/10.5962/bhl.title.68927. So similar and yet so different: taxonomic status of Pallid Swift Apus pallidus Lister, A., 1989. Rapid dwarfing of red deer on Jersey in the last interglacial. Nature and Common Swift Apus apus. Hous. Theor. Soc. 64, 344e352. 342, 539e542. https://doi.org/10.1038/342539a0. Phillimore, A.B., Owens, I.P., 2006. Are subspecies useful in evolutionary and con- Maddison, W.P., Maddison, D.R., 2018. Mesquite: a Modular System for Evolutionary servation biology? Proc. R. Soc. B 273, 1049e1053. https://doi.org/10.1098/ Analysis. Version 3, 51 ed. rspb.2005.3425. Martin, T.E., Kelly, D., Keogh, N., Heriyadi, D., Singer, H., Blackburn, G., 2012. The Phillimore, A.B., Owens, I.P.F., Black, R.A., Chittock, J., Burke, T., Clegg, S.M., 2008. avifauna of lambusango forest reserve, Buton island, south-east Sulawesi, with Complex patterns of genetic and phenotypic divergence in an island bird and additional sightings from southern Buton. Forktail 28, 107e112. the consequences for delimiting conservation units. Mol. Ecol. 17, 2839e2853. Martin, T.E., Kelly, S.B.A., Froese, G., Brodie, J., Mulyani, Y., Kelly, D.J., 2015. New https://doi.org/10.1111/j.1365-294X.2008.03794.x. records from Buton island, Sulawesi, Indonesia. BirdingASIA 23, 128. Puillandre, N., Lambert, A., Brouillet, S., Achaz, G., 2012. ABGD, automatic barcode Martin, T.E., Monkhouse, J., Akbar, P.G., Baddams, J., Mulyani, Y., Kaban, A., et al., gap Discovery for primary species delimitation. Mol. Ecol. 21, 1864e1877. 2017. Notes on the distribution, status and natural history of ten species on https://doi.org/10.1111/j.1365-294X.2011.05239.x. Buton island, Southeast Sulawesi, Indonesia. BirdingASIA 74e77. R Core Team, 2020. R: a Language and Environment for Statistical Computing. Mayr, E., 1942. Systematics and the Origin of Species. Columbia University Press, version 4.0, 2 ed. R Foundation for Statistical Computing, Vienna, Austria. New York. Rambaut, A., Drummond, A.J., Xie, D., Baele, G., Suchard, M.A., 2018. Posterior Mayr, E., 1959. Isolation as an evolutionary factor. Proc. Am. Phil. Soc. 103, 221e230. summarization in Bayesian phylogenetics using Tracer 1.7. Syst. Biol. 67, Mayr, E., 1963. Animal Species and Evolution. Harvard University Press, Cambridge, 901e904. https://www.ncbi.nlm.nih.gov/pubmed/29718447. Massachusetts. Redfern, C.P.F., Clark, J.A., 2001. Ringers' Manual. British Trust for Ornithology, Mayr, E., 1969. Bird speciation in the tropics. Biol. J. Linn. Soc. 1, 1e17. https:// Thetford. doi.org/10.1111/j.1095-8312.1969.tb01808.x. Rentschlar, K.A., Miller, A.E., Lauck, K.S., Rodiansyah, M., Bobby, Muflihati, et al., McDonald, P.G., Griffith, S.C., 2011. To pluck or not to pluck: the hidden ethical and 2018. A silent morning: the songbird trade in Kalimantan, Indonesia. Trop. scientific costs of relying on feathers as a primary source of DNA. J. Avian Biol. Conserv. Sci. 11, 1940082917753909 https://doi.org/10.1177/ 42, 197e203. https://doi.org/10.1111/j.1600-048X.2011.05365.x. 1940082917753909.

324 F. O Marcaigh, D.J. Kelly, D.P. O'Connell et al. Zoologischer Anzeiger 293 (2021) 314e325

Rheindt, F.E., Prawiradilaga, D.M., Ashari, H., Suparno, Gwee, C.Y., Lee, G.W.X., et al., Toews, D.P., Brelsford, A., 2012. The biogeography of mitochondrial and nuclear 2020. A lost world in Wallacea: description of a montane archipelagic avifauna. discordance in animals. Mol. Ecol. 21, 3907e3930. https://www.ncbi.nlm.nih. Science 367, 167e170. gov/pubmed/22738314. Robinson-Dean, J.C., Willmott, K.R., Catterali, M.J., Kelly, D.J., Whittington, A., Torchiano, M., 2020. Package "effsize": Efficient Effect Size Computation. CRAN Phalan, B., et al., 2002. A new subspecies of Red-backed Thrush Zoothera Repository 1-10. https://doi.org/10.5281/zenodo.196082. erythronota kabaena subsp. nov. (Muscicapidae: turdidae) from Kabaena island, Trethowan, L.A., Eiserhardt, W.L., Girmansyah, D., Kintamani, E., Utteridge, T.M.A., Indonesia. Forktail 1e10. Brearley, F.Q., 2020. Floristics of forests across low nutrient soils in Sulawesi, Rubinoff, D., Holland, B.S., 2005. Between two extremes: mitochondrial DNA is Indonesia. Biotropica 1e10. https://doi.org/10.1111/btp.12838. neither the panacea nor the nemesis of phylogenetic and taxonomic inference. Trier, C.N., Hermansen, J.S., Sætre, G.-P., Bailey, R.I., 2014. Evidence for mito-nuclear Syst. Biol. 54, 952e961. https://doi.org/10.1080/10635150500234674. and sex-linked reproductive barriers between the hybrid Italian sparrow and its Ryder, O.A., 1986. Species conservation and systematics: the dilemma of subspecies. parent species. PLoS Genet. 10, e1004075 https://doi.org/10.1371/ Trends Ecol. Evol. 1, 9e10. https://doi.org/10.1016/0169-5347(86)90059-5. journal.pgen.1004075. Satyana, A.H., Purwaningsih, M.E.M., 2011. Collision of Micro-continents with Tweedley, J., Bird, D., Potter, I., Gill, H., Miller, P., O'Donovan, G., et al., 2013. Species Eastern Sulawesi: Records from Uplifted Reef Terraces and Proven-Potential compositions and ecology of the riverine ichthyofaunas in two Sulawesian Petroleum Plays. Proceedings of the Indonesian Petroleum Association. islands in the biodiversity hotspot of Wallacea. J. Fish. Biol. 82, 1916e1950. Sclater, P.L., 1859. Extracts from correspondence, notices, etc. Ibis 1, 111e116, https://doi.org/10.1111/jfb.12121. 10.1111/j.1474-919X.1859.tb06192.x. United Nations, 1992. Convention on Biological Diversity. Rio de Janeiro. Smith, M.L., Carstens, B.C., 2020. Process-based species delimitation leads to iden- Vartanyan, S., Garutt, V., Sher, A.V., 1993. Holocene dwarf mammoths from wrangel tification of more biologically relevant species. Evolution 74, 216e229. https:// island in the siberian arctic. Nature 362, 337e340. https://doi.org/10.1038/ doi.org/10.1111/evo.13878. 362337a0. Smith, B.T., McCormack, J.E., Cuervo, A.M., Hickerson, M.J., Aleixo, A., Cadena, C.D., Vigors, N.A., Horsfield, T., 1826. IX. A description of the Australian birds in the et al., 2014. The drivers of tropical speciation. Nature 515, 406e409. https:// collection of the Linnean Society; with an attempt at arranging them according doi.org/10.1038/nature13687. to their natural affinities. Transact. Linnean Soc. London 15, 170e331. https:// Stattersfield, A.J., Crosby, M.J., Long, A.J., Wege, D.C., 1998. Endemic Bird Areas of the doi.org/10.1111/j.1095-8339.1826.tb00115.x. World: Priorities for Biodiversity Conservation. BirdLife International, Cam- Voous, K.H., 1952. A new race of babbler from the togian islands, North celebes. bridge, UK. Ardea 40, 74. Stresemann, E., 1931. Neue Vogelrassen aus dem Latimodjong-Gebirge. Ornithol. Walden, A., 1876. Description of a new species of the genus Trichostoma from the Monatsber. 39, 7e14. island of Celebes. Ibis 18, 376e379. https://doi.org/10.1111/j.1474- Stresemann, E., 1938. Letzte Erganzungen€ zur Avifauna von Celebes. Ornithol. 919X.1876.tb06933.x. Monatsber. 46, 147e149. Wallace, A.R., 1880. Island Life, or the Phenomena and Causes of Insular Faunas and Strickland, H.E., 1849. Trichostoma celebense, Strickland. Contributions to orni- Floras. Macmillan and Co., London. thology for 1849 127-28. https://www.biodiversitylibrary.org/page/51703112. Wallace, A.R., 1887. Oceanic islands: their physical and biological relations. J. Am. Swainson, W., 1831. Appendix, No. I. Characters of genera and sub-genera hitherto Geogr. Soc. N. Y. 19, 1e21. https://doi.org/10.2307/196724. undefined. In: Richardson, J., Swainson, W., Kirby, W. (Eds.), Fauna Boreali- White, C., Bruce, M., 1986. The Birds of Wallacea (Sulawesi, the Moluccas & Lesser Americana, or, the Zoology of the Northern Parts of British America. John Sunda Islands, Indonesia). British Ornithologists' Union, London. Murray, London, pp. 481e497. https://doi.org/10.5962/bhl.title.39293. Whittaker, R.J., Fernandez-Palacios, J.M., Matthews, T.J., Borregaard, M.K., Templeton, A.R., Crandall, K.A., Sing, C.F., 1992. A cladistic analysis of phenotypic Triantis, K.A., 2017. Island biogeography: taking the long view of nature's lab- associations with haplotypes inferred from restriction endonuclease mapping oratories. Science 357. https://doi.org/10.1126/science.aam8326. and DNA sequence data. III. Cladogram estimation. Genetics 132, 619e633. Wiens, J.A., Mayr, E., Parkes, K.C., Gill, F.B., Storer, R.W., Barrowclough, G., et al., 1982. The Geneland Development Group, 2020. Population genetic and morphometric Forum: avian subspecies in the 1980's. Auk 99, 593e615. https://doi.org/ data analysis using R and the Geneland program. In: Guillot, G. (Ed.), Dardilly, 10.1093/auk/99.3.593. France. International Prevention Research Institute. Wilson, E.O., Brown, W.L., Jnr, 1953. The subspecies concept and its taxonomic Tobias, J.A., Seddon, N., Spottiswoode, C.N., Pilgrim, J.D., Fishpool, L.D.C., Collar, N.J., application. Syst. Zool. 2, 97e111. https://doi.org/10.2307/2411818. 2010. Quantitative criteria for species delimitation. Ibis 152, 724e746. https:// Yu, G., Smith, D.K., Zhu, H., Guan, Y., Lam, T.T.Y., McInerny, G., 2016. GGTREE: an R doi.org/10.1111/j.1474-919X.2010.01051.x. package for visualization and annotation of phylogenetic trees with their Tobias, J.A., Ottenburghs, J., Pigot, A.L., 2020. Avian diversity: speciation, macro- covariates and other associated data. Methods Ecol. Evol. 8, 28e36. https:// evolution, and ecological function. Annu. Rev. Ecol. Evol. Syst. 533e560. https:// doi.org/10.1111/2041-210X.12628. doi.org/10.1146/annurev-ecolsys-110218-025023.

325